Flash flooding in Spain: geomorphological approaches supporting flood frequency analysis, and the implications for the design of structures

Flash flooding in Spain: geomorphological approachessupporting flood frequency analysis, and the implicationsfor the design of structures

Alma Lopez-Aviles MCIWEM

Development Control Engineer/Climate Change Policy Advisor, Environment Agency

Keywords

dating; flash flooding; flood frequency; fluvial

deposits; geomorphological; hydraulic;

hydrological; return period.

Correspondence

Alma Lopez-Aviles, Development

Control Engineer/Climate Change Policy

Advisor, Environment Agency, Swift House,

Frimley Business Park, Firmley, Camberley,

Surrey, GU16 7SQ, UK.

Email: alopez_aviles@environment-agency.

gov.uk

CIWEM 2005 Young Author’s Competition, 18

May 2005, Imperial College, London, UK.

doi:10.1111/j.1747-6593.2006.00063.x

Abstract

Conventional methods for the estimation of flood frequency are generally

based on the statistical analysis of data series resulting from the measurement

of water levels at specific locations, which are translated into discharges (m3/s)

using standard stage/discharge relationships. Subsequently, these gauged flows

are used to identify the largest flood event experienced by a river or catchment,

and to produce growth curves used in the estimation of the return periods of

specific flood events, as well as in the calculation of discharges for specific

required design events (e.g. return period 1 in 100 year). In areas where

gauging data records are scarce, and/or the data series are short or interrupted,

geomorphological interpretation of the physical environment, dating of fluvial

deposits and flooding episodes and hydraulic reconstruction of past flood

events can be used as complementary tools aiding conventional hydrological

and flood frequency analysis methods. This paper will discuss the undertaking

of this approach in the Guadalope Catchment in northeast Spain. It will

examine the findings in relation to the inadequate design of existing structures

such as dams, spillways, canals and reservoirs, and will also look at the potential

risks associated with flooding at present.

Introduction

There are more than 1200 large (4 15 m high) dams in

Spain, which generate 12% of the country’s electricity

(World Commission on Dams 2000). Furthermore, the

supply of water to towns and villages is largely dependent

on water stored in reservoirs built behind these dams.

Like in many other parts of the world, in Spain,

gauging and rainfall data series are often short or incom-

plete, especially in the more remote mountainous areas

where large dams and reservoirs are located. Further-

more, catchment characteristics and the typically Medi-

terranean rainfall pattern in many regions of eastern and

southern Spain make the prediction of flooding events of

specific return periods a difficult task. A combination of

steep slopes, low vegetation density, narrow bedrock

channels and the antecedent conditions of many catch-

ments, together with high-intensity localised rainfall

episodes lead to high-velocity large-magnitude flash

flooding along many Mediterranean basins.

In addition to gauged data analysis, unconventional

approaches to flood frequency estimation can help in the

first stages of the design of structures such as large dams

by improving safety factors. The following sections of this

paper will describe the research methods used in the

Guadalope Catchment, and will discuss the main findings

in relation to the risks associated with the poor safety

factors of recently constructed dams.

Research area and methods

The Guadalope Catchment

The Guadalope River flows northward from the uplands

of El Maestrazgo in northeast Spain to the Ebro River

basin. The main tributary of the Guadalope is the Ber-

gantes River, which joins in from the southeast part of the

catchment (see Fig. 1). The drainage area of the Guada-

lope basin is 3892 km2, with the Bergantes sub-basin

accounting for 1221 km2 of the total area. The upland

part of the catchment is set on highlands of the Iberian

Range, where the topography, adverse climate, precarious

communications infrastructure and lack of employment

Water and Environment Journal 21 (2007) 217–226 c� 2007 The Author. Journal compilation c� 2007 CIWEM. 217

Water and Environment Journal. Print ISSN 1747-6585

have driven local populations away towards more eco-

nomically developed parts of the region or country.

By European standards, the density of population in

Spain is quite low (i.e. approximately 40 million people

live in the second largest country in Western Europe, thus

approximately 79 inhabitants/km2). Spain is also consid-

ered to be the second most mountainous country in

Europe (World Bank 2004). Some areas of the Iberian

Range are among the less populated and less developed

parts of the country. It is in these mountainous areas that

many dams and reservoirs have been built, and where

gauging and rainfall data are scarce and series are often

incomplete.

Methods

Given the few gauging stations available in the catch-

ment (i.e. Station 30 on the upstream part of the Guada-

lope, Station 31 on the Bergantes, Station 15 on the

Guadalope in Alcaniz and Station 99 on the Guadalope

near its confluence with the Ebro River in Caspe – see

Fig. 1), and the short length and poor quality of

the gauging records, it was considered that a compilation

of historic information on flood events would help

to build the scenario regarding flooding within the

catchment.

Flood History Databases (Comision Tecnica de Inunda-

ciones 1985) for the region as well as historic documents

held in local libraries were consulted in order to obtain

references about exceptionally large rainfall or flood

events experienced within the Guadalope Catchment.

These references were recorded unsystematically in a

variety of documents, and therefore it is acknowledged

that a greater number of significant flood events might

have occurred within this basin, but could have been

missed in the search. Furthermore, interviews with el-

derly members of local communities who had witnessed

recent flood events were conducted to provide supple-

mentary information to historic documents. This compi-

lation of historic information on past flood events helped

Reservoirs

Meteorologicalstations

Gauging stations

Settlements

Zaragoza

Mequinenza

Riba-Roja

40°N

5°W

0 250km

Calanda

Study Site

Castelserás

Alcaniz~

N

0 10 20 30 km

C a

t a

l a

n i

d e

s

Morella Vella

Morella

Castellfort

IglesuelaFortanete

Station 88

V. Pinares

C. Almuden AliagaM. Jarque

Station 100

Station 201

Station 106

Station 15

Station 99

Caspe

Caspe

Valmuel

E. Estanca

Andorra

Station 30Station 31Santolea

CastelloteM. Matas

Torrevelilla

Cerollera

Aguaviva

GallipuenCalanda

IberianRange

Pyrenees

Ebro River

A

.R

ega l

loB

aj o

Ar a g ó n

Mar

tín R

iver

Aguas

Vivas River

Mat

arra

na

Riv

er

Mediterranean SeaMediterranean SeaS i e r r a d e l M a e s t r a z g o

Gua

dalo

peRiv

er

Ebro

Bergan

t es River

~

Mediterranean Sea

Fig. 1. Location map of the Guadalope Catchment, including gauging and rainfall stations. A. Regallo, Arroyo Regallo; C. Almuden, Cuevas de Almuden;

E. Estanca, Embalse de la Estanca; M. Jarque, Mezquita de Jarque; M. Matas, Mas de las Matas; V. Pinares, Villarroya de Pinares.

Water and Environment Journal 21 (2007) 217–226 c� 2007 The Author. Journal compilation c� 2007 CIWEM.218

Flash flooding in Spain A. Lopez-Aviles

in the flood frequency analysis process by extending the

data series.

Simultaneously, a bedrock-controlled reach of the Ber-

gantes River (see Fig. 1) was selected as representative of

the characteristics of this subcatchment, and thus where

detailed geomorphological analysis, hydraulic reconstruc-

tion and dating of deposits were to take place. The

characteristics of this study area and its sedimentation

regime are described in more detail in Lopez-Aviles et al.

(1998) and Lopez-Aviles (1998).

A number of geomorphological units (e.g. fine deposits,

gravel terraces, bars, bedrock benches) were identified,

and sediment samples were taken for dating of the

respective units, aiming to interpret incision and aggrada-

tion cycles in relation to dominant climatic conditions.

Sediment dating techniques included: (a) 14C analysis of

organic matter contained within slack-water deposits,

(b) dendrochronology of trees growing on the floodplain

(i.e. establishing the age of these trees assumes a mini-

mum age for relevant floodplain sedimentary gravel

units, and therefore a minimum period since the last time

a large flood event affected the specific units) and (c)

luminescence (infrared-stimulated luminescence – IRSL)

dating of sediment grains (mainly feldspar and quartz)

showing the last time that these deposits were exposed to

light before being buried by a flood. Luminescence meth-

ods measure the radiation accumulated in the minerals

once they were buried.

The study site was surveyed and mapped in detail (see

Fig. 2), and hydraulic reconstruction of flood conditions

derived from high water marks (slack water deposits,

debris) at specific locations was undertaken at relevant

cross-sections (see Fig. 3). The purpose of this exercise

was to estimate the order of magnitude of known flood

events that either predated gauging stations or were not

measured at the existing ones, but for which historic and/

or geomorphological evidence was available.

The hydraulic reconstruction of past flood events in-

volved the calculation of flow velocities and discharges

using standard formulae (i.e. continuity equation with

Manning’s velocity and roughness coefficient):

Q ¼ wdV ; ð1Þ

where Q is the discharge in m3/s, V is the mean velocity, w

is the mean channel width and d is the mean depth equal

to the cross-sectional area divided by bankful width:

V ¼ R2=3S1=2=n; ð2Þ

where V is the mean velocity in m/s, R is the hydraulic

radius equal to the cross-sectional area divided by the

wetted perimeter, S is the average water surface slope and

n is the Manning’s roughness coefficient. This was calcu-

lated for the various locations using a number of methods.

For further details, see Chapter 5 in Lopez-Aviles (1998).

Velocities and flow discharges were calculated at spe-

cific cross-sections to match the level of dated sediment

deposits or high water marks as illustrated in Fig. 3.

Therefore, detailed surveying of channel slope (long

profile) and cross-sections of the channel and floodplain,

as well as grain-size analysis for the estimation of rough-

ness coefficients, were undertaken (see Figs 2 and 3).

This approach to reconstruct the hydraulic parameters

involves assuming that aspects such as channel geometry

and roughness have not changed significantly over time.

This is believed to be the case for the time period studied,

given that the reach described in this paper is a bedrock-

controlled reach, where major modifications to the chan-

nel geometry would necessitate longer geological times.

Furthermore, it must be emphasised that the intrinsic

uncertainties associated with this assumption would be

significant for very detailed modelling and/or design in

the studied reach, but they are considered acceptable for

the purpose of this study, which is to estimate the order of

magnitude of past flood events.

Heights (m)

40–4546–5051–5556–6061–6566–70

>70Flow direction

0 100 m

N

Topography

Fig. 2. Geomorphological units identified and surveyed at the study site.


Flash flooding in SpainA. Lopez-Aviles

Discussion

Hydrology of the Guadalope Catchment

The available rainfall data collected in various stations

across the Guadalope Catchment (see Fig. 1) were ana-

lysed and compared spatially. Figure 4 illustrates the poor

quality of data with numerous interruptions to the data

sets (Fig. 4, see graphs a0 and b0), and the irregularity,

both in time and space, of the rainfall across the catch-

ment. Graph a–a0 represent rainfall patterns in the down-

stream part of the catchment, whereas graph b–b0

illustrates the situation for the upstream highlands. Over-

all, the data available show great disparity in terms of the

total rainfall in different years. The interannual differ-

ences in the monthly mean recorded are also obvious,

which indicate a highly irregular pattern of precipitation

in the area and suggest that sporadic, possibly infrequent

but highly intense rainfall episodes occur some years

that affect the monthly mean rainfall recorded. This has

important repercussions on the flooding patterns of the

Guadalope and Bergantes Rivers.

Spatially, the catchment is characterised by a dual

pattern of precipitation, with the Bergantes subcatchment

being most affected by an Autumn peak due to highly

seasonal precipitation, and the Guadalope subcatchment

experiencing a main annual peak in Spring and a second

peak in Autumn.

Flooding pattern in the Guadalope Catchment

The very intense and localised nature of some rainfall

episodes experienced in the area, especially in the Au-

tumn, together with the steep slopes, sparse vegetation

and impermeable bedrock substrate, give rise to very

sudden increases in flow, which can go from less than

10 m3/s to more than 650 m3/s in 24 h, and that can reach

instantaneous peak discharges above 1000 m3/s.

This appears to have been the case in October 1967,

where a sudden ‘flash flood’ was recorded at Stations 31

Cross-section 9

Cross-section 8

Cross-section 6Cross-section 7

Cross-section 5

Cross-section 4

Cross-section 3

Cross-section 2

Cross-section 1N

0 100 m

1

2

3

4

5

6

8

970

65

60

55

50

45

400 50 100 150 200 250

Hei

ght (

m)

Channel width (m)

70

65

60

55

50

45

400 50 100 150 200 250

Hei

ght (

m)

Channel width (m)

Cross section 7 at reach C and location of 14Cand IRSL sample 2.

Cross section 4 at reach A and locationof IRSL sample 3.

Cross-section 7

Cross-section 4

IRSL sample 2

IRSL sample 3

14Csample

7

GED

Gravel units on floodplain

F

(a) (b)

(c)

Fig. 3. Cross-sections along the studied reach of the Bergantes River (a), and detailed sections showing the levels of sediment samples dated (right).

IRSL, infrared-stimulated luminescence.



and 15 on the Bergantes and Guadalope Rivers, respectively

(see Fig. 5). Other flash-flood events have been recorded in

the catchment, but the highest instantaneous peak dis-

charge measured in the catchment was that of 1560 m3/s

recorded on the Bergantes River in October 1967.

The magnitude and unpredictable nature of these sud-

den events highlights the problems faced, not only in terms

of and the lack or shortness of flood warning periods, and

therefore the catastrophic effects that can be experienced,

but also in terms of the accuracy of the records, mainte-

nance and operation of gauging equipment, and therefore,

the difficulties in maintaining a continuous record.

The flash flood experienced in Mas de las Matas in

August 1996 (see Fig. 6) illustrates the problems described

above. No warning was available for this localised event

that affected the village’s centre and numerous infrastruc-

ture, as well as the local economy (i.e. the annual fruit

harvest was lost). The flood event was not recorded as the

stream running through the village is only a small un-

gauged tributary of the Guadalope River.

The difficulties in estimating potential high flows and

their return period based on gauged data have resulted in

great uncertainties when assessing the risk of flooding for

existing settlements, and more importantly, when plan-

ning and building new infrastructure. Therefore, it is

believed that the risks have been underestimated and

Monthly mean rainfall at Alcaniz(1929–34/1958–76/1983–94)

0255075

100125150175200

Jan

Feb

Mar

Apr

May Jun

Jul

Aug

Sep Oct

Nov

Dec

Jan

Feb

Mar

Apr

May Jun

Jul

Aug

Sep Oct

Nov

Dec

Pre

cipi

tatio

n (m

m)

Pre

cipi

tatio

n (m

m)

MaximumAverage ppMinimum

Total annual rainfall at Alcaniz (1929–34/1958–76/1984–86/1988–94)

0100200300400500600700

1929

1933

1937

1941

1945

1949

1953

1957

1961

1965

1969

1973

1977

1981

1985

1989

1993

Monthly mean rainfall at Castellfort (1929–32/1946–71/1994)

0255075

100125150175200225250

Pre

cipi

tatio

n (m

m)

MaximumAverage ppMinimum

Total annual rainfall at Castellfort (1930/1932/1950–52/1955–71/1994)

0

200

400

600

800

1000

1930

1934

1938

1942

1946

1950

1954

1958

1962

1966

1970

1974

1978

1982

1986

1990

1994

Pre

cipi

tatio

n (m

m)

(a)

(b)

(a')

(b')

Fig. 4. Monthly mean rainfall and annual rainfall data series recorded in Alcaniz (downstream in the Guadalope Catchment) and Castellfort (uplands).

700

Catastrophic flood recorded in the Bergantes Riverin October 1967, gauged at station 31

Catastrophic flood recorded in the Guadalope Riverin October 1967, gauged at station 15

600500400300200100

0

700600500400300200100

0

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

Days

Days

Dis

char

ge (

m3

s−1)

Dis

char

ge (

m3

s−1)

(a)

(b)

Fig. 5. Flood event of October 1967 recorded on the Bergantes River

(station 31) see (a), and the Guadalope River in Alcaniz (station 15)

(downstream part of the catchment) see (b); bars represent mean daily

discharges recorded at gauging stations 31 and 15.



poor safety factors have been included in the design of

some structures as described in the following section.

These difficulties have also been transferred to policy

making and drawing regulations in the area.

Dams and reservoirs in the Guadalope Catchment

Table 1 illustrates the specifications of the most important

infrastructure (dams and reservoirs) regulating the flow

regime in the Guadalope Catchment. Locations are shown

in the area map: Fig. 1.

As illustrated in the above table, the outlet capacity of

the Calanda dam’s spillway is smaller than some of the

registered flows in the upland part of the catchment

(i.e. 1560 m3/s recorded at the confluence of the rivers

Bergantes and Guadalope). This suggests that potential

risks exist that are associated with the possibility of high-

magnitude flood events re-occurring in the catchment.

On such occasions, the outlet capacity of the Calanda

dam’s spillway might not be able to accommodate the

volume of flow necessary to empty the reservoir in time.

This could lead to failure or overflowing of the Calanda

Fig. 6. Sequence shows a dry channel through Mas de las Matas, and the flash flood of August 1996 at the same location, plus postevent damage on

the site.

Table 1 Specifications of dams and reservoirs in the Guadalope Catchment

Name of dam

and reservoir Location

Reservoir storage

capacity (Hm3)

Maximum outlet

discharge at

spillway (m3/s)

Designed/modelled

maximum discharge

(m3/s)

Maximum

instantaneous

discharge

recorded (m3/s)

Gallipuen Upstream on tributary of Guadalope 3.5 250 � �

Santolea Upstream on Guadalope 54 1156 � �

Calanda Confluence of Bergantes and Guadalope 54 1425 1250 1560

Caspe Downstream on Guadalope 78 1560 � �

Mequinenza On Ebro at confluence with Guadalope 1530 � � �

�Data not available.

Hm3 is cubic hectometres (1 Hm3 =1 000 000 cubic metres).



dam, and to a situation similar to that at the Gallipuen

and/or Santolea dams, upstream, with the worse-case

scenario being the possibility of a break ‘in cascade’ of

these dams.

The Caspe dam was designed with this in mind and

having been constructed at a later date, the spillway

capacity can accommodate flows comparable to the ones

registered in the 1967 event. Furthermore, in the case of

the Calanda dam and reservoir, this structure is located

immediately downstream from the confluence between

the Guadalope and Bergantes Rivers (see Fig. 1). This

makes the Calanda dam the most vulnerable of the

catchment, as it can be directly influenced by extreme

flood events from either tributary. Additionally, unlike

the Caspe reservoir, the Calanda dam is located in the

upland part of the catchment where gradients are still

fairly steep, which means flow velocities here can be high.

These factors, together with the unpredictability of

storm events that can lead to localised flooding in the

region, have led the relevant authorities to operate the

Calanda reservoir at a lower than designed capacity at all

times in order to compensate for the design’s poor safety

factors. This decision has been taken using the precau-

tionary principle and in trying to minimise the risks to the

downstream areas of the catchment in the event of failure

or overtopping of this dam.

Dating of sediment deposits and hydraulicreconstruction

Dates obtained from the dating of fine sediment deposits

using luminescence (IRSL), as well as 14C techniques are

illustrated in Table 2. Luminescence methods have also

been used to date Quaternary deposits in this catchment

by Fuller et al. (1996). The resulting dates from this

research were compared and are in agreement with those

previously obtained in the Guadalope Catchment.

The dates obtained for these deposits include a margin

of error that can be substantially large for older Pleisto-

cene deposits. This reflects some of the intrinsic problems

associated with dating old materials. There are other

possible sources of error with Luminescence techniques

as a number of assumptions have to be made, such as that

the flood event occurred during day-time clear waters (in

order for date to reflect the last time, sediments were

exposed to light), and that no subsequent scouring or

minor reworking of material has taken place (see Lopez-

Aviles et al. (1998) and Lopez-Aviles (1998) for more

details).

The 14C date illustrated in Table 2 provides a maximum

age of deposition for these sediments, and reflects the

effects of the period when 14C atmospheric content

increased rapidly as a result of widespread testing of

nuclear devices. Peak concentrations were recorded in

1963; thus, the two possible dates shown in Table 2

indicate 14C content on either side of the curve for which

the peak is placed around 1963.

Despite the uncertainties, it is considered that these

dates provide additional useful information about the

flooding mechanisms, flooding intensity and, to some

degree, the regularity of high-magnitude flooding in this

region. In particular, this is the case with the two most

recent dates. These suggest that flood conditions were of

such magnitude sometime between 1395 and 1545, and

again in the 1950s or 1980s, as to have resulted in

sediments being deposited at heights between 10 and 8 m

above the current bedrock channel bed. Reconstruction of

the hydraulic conditions in these situations was under-

taken in order to estimate the order of magnitude of the

flows necessary to have deposited fine sediments up to

the described levels (see Fig. 3 and the method described

in ‘Methods’). The results of this exercise are shown in

Table 3.

The following four points summarise the interpretation

of results obtained from the hydraulic reconstruction:

(a) Very high-magnitude flood events have been experi-

enced in the Bergantes River, and thus within the Guada-

lope Catchment, in the recent past. This is corroborated by

gauged data (e.g. 1967 event), and historic, field and

photographic evidence.

(b) For flood water levels to reach the level of the Island

Unit (see Fig. 3, bottom right) while the two channels

Table 2 Sediment samples location and dates provided by infra-red stimulated luminescence (IRSL) and 14C techniques

Unit Type of sediment material

Location in metres above

the current channel bed

Dates in thousands of years before present

(ka BP) for older dates, AD for most recent dates

Technique

used

Pleistocene terrace Gravels and fines 4 37.8� 4.2 ka BP IRSL

Holocene fill terrace Fine sediment 7 9.6� 1.6 ka BP IRSL

Fine deposits on Island

unit

Fine deposits embedded on

bedrock substrate

10 Dated between 1395 and 1545 AD IRSL

Slack-water deposits Fines deposited in alcove

within a gravel terrace

8 Dated late 1950s or late 1980s AD14C

Pleistocene: 1.8 million to 11 000 years ago and Holocene: last 10 000 years; AD (Anno Domini) to signify the Christian era: years after Christ was born.

BP refers to years before present, this being 1950.



around the island are flowing at full capacity, extremely

large discharges (up to 7500 m3/s) are required. This is

considered less likely than the alternative, which could

involve the secondary channel being obstructed by debris,

thus blocking flow passage through this channel during

peak flows, and rerouting the flow over the Island unit.

(c) The discharge obtained for the slack water sediments

is thought to overestimate the flow necessary to deposit

material in an alcove that is located on the outside of a

sharp (901) meander (see Fig. 7). This is due to the fact

that uniform steady flows across the cross-section at this

location are unlikely.

(d) The gravels and boulders laying on the current flood-

plain (see Fig. 3c) are reworked regularly as indicated by

that lack of any vegetation, lichens, etc., which suggests

that discharges between 30 and 250 m3/s are experienced

in the Bergantes subcatchment frequently.

Recurrence interval (RI) of flood events

Following the exercise of estimating discharges for past

flood events, and as stated in previous sections, all historic

information as well as field evidence of past flood events

within the Guadalope Catchment was used to extend the

catchment’s flood data series. Subsequently, a basic calcu-

lation of the (RI for flood events above specific magni-

tudes was undertaken using the following equation as

described in Bull (1998):

RI ¼ nþ 1=m; ð3Þ

where n is the number of years of record and m is the

rank, 1 for the largest flood.

The approach is based on standard statistical flood

frequency analysis methods, which vary according to the

flood distribution graphical representation, but that are

based on a function of T is the return period or RI, n is the

number of values and m or r are the rank from top. In this

case, the flood-frequency approach used was only applied

to the identified dated flood events that fell within a

period of time considered as ‘historic’ comprising the last

200 years as a maximum, and not to older geomorpholo-

gical events. This is due to the fact that the method

assumes a continuous data series for a determined num-

ber of years (n), which is not strictly accurate for anecdo-

tal evidence – whether documented or field-based.

Large flood events examined in this paper were as-

sumed to have been above 100–150 m3/s for them to have

been recorded as ‘extraordinary’ in historical sources,

testimonies and in the field. Even though the events

presented in this paper were not recorded systematically,

it has been considered that it is acceptable to assume a

continuous data series for the unusually large flood

events experienced during historic times. However, many

more events might have occurred in the area, for which

information has been missed or was not recorded. There-

fore, it must be noted that the RIs estimated in this paper

represent a minimum return period for historic large flood

events.

Fig. 7. Location of alcove (a) and slack-water deposits (b) dated by 14C. For reference the river flows from top to bottom (901) in the far left part of (a).

The white dot to the centre-right part (a) is a person standing outside the alcove.

Table 3 Discharge estimates obtained via hydraulic reconstruction

Unit name

Discharge

estimates in m3/s

Holocene fill terrace dated 9.6� 1.6 ka BP �2000

Fine sediments on Island unit dated

1395–1545 AD

�7500

Slack-water deposits dated of 1950s or

1980s AD

�3600

Gravel units on current floodplain 30–250

ka BP for thousands of years before present (1950).

AD, Anno Domini.



However, the further back into the past (i.e. for geo-

morphological events), the larger the uncertainty about

whether flood events for which evidence/records are not

available might have occurred in between known flood-

ing episodes. Therefore, the data series cannot be assumed

as continuous, and thus, no attempt has been made in this

paper to analyse the possible return period for these

geomorphological events, as the results could be mislead-

ing. The data associated with older geomorphological

events shown in this paper are presented as further

evidence that large catastrophic flood events do occur in

the Guadalope–Bergantes Catchment.

The method used in this research shows how to extend

the length of conventional gauged data series by adding

historic and field evidence of past large flood events. It

also illustrates how it can be useful to investigate historic

evidence of large-magnitude floods before designing

structures, especially if the available gauging and rainfall

records are not very long. However, the results have to be

treated with caution and can only be seen as an attempt

to estimate minimum RIs for historic flood events based

on a series of assumptions.

The results from this exercise as well as the main

conclusions of this research are described further in the

following sections.

Summary of results

The main conclusions from this research indicate that

very high-magnitude flood events have been experienced

by the Bergantes River in historic times. This type of

flooding is believed to be more frequent than previously

thought in this catchment. This has been corroborated by

gauged data recorded both on the Guadalope and Ber-

gantes Rivers in the last few decades, and by the rainfall

pattern of the region, which is characterised by sporadic

localised and very intense rainfall events.

Regarding the functional floodplain of the Bergantes

River, estimates of past discharges calculated using stan-

dard formulae (see ‘Methods’) suggest that discharges

between 30 and 250 m3/s are required to rework the

active floodplain. By comparing these magnitudes with

gauged data for the Bergantes River, it is estimated that

reworking of the active floodplain takes place every year

or every other year.

Following basic analysis of RIs for flood events of

specific magnitudes (see Eq. 3), the 1967 event (known

to be above 1500 m3/s according to gauging station in-

formation) was ranked alongside known events recorded

for this area as far back as 1945 (approximately 1000 m3/s

downstream from the study reach). Based on the data

available, magnitudes of a similar order to that of the 1967

event are believed to have a minimum return period of

1 in 50 years in the Bergantes River. However, if the 14C

date of the 1950s or the 1980s – obtained from deposits

located within the study site – is included in the frequency

analysis, the suggestion is that flood events of an order of

magnitude larger than 1000–1500 m3/s can occur even

more frequently in the Bergantes Catchment (i.e. RI up to

1 in 25 years). This is corroborated by geomorphological

evidence of very large flood events that have occurred in

the Bergantes Catchment in the past (see Tables 2 and 3).

Conclusions

(1) Given that large flash-flooding episodes appear to be

more common in the Bergantes Catchment than pre-

viously thought, and as these flood events seem to be

responsible for most of the geomorphological work in the

valleys, it is recommended that any new structures

planned within the Guadalope–Bergantes Catchment

should be designed accordingly. Inadequate design of

spillways for dams and reservoirs may potentially lead to

failure or under-capacity to empty reservoirs, which may

in turn have catastrophic effects on the local infrastruc-

ture and economy. The area’s transport system including

roads, railways and bridges runs parallel and across water-

courses, especially in the most mountainous upper part of

the catchment. Additionally, the cooling system of the

thermal power station of Andorra depends on water

supplied from the Calanda reservoir via a canal that flows

parallel to the Guadalope River. Therefore, the lack of

water supply caused by damage to either the canal or

reservoir would have significant repercussions.

(2) Flooding by failure or overtopping of dams, and the

potential risk of a break of dams ‘in cascade’ would also

have important repercussions for the local economy. The

damages would be derived from direct impact on the area’s

agriculture, farming and mining (coal) activities, and

indirectly, due to the transport disruptions caused. The

potential damages derived from loss of life, and the

devastating effects on human settlements are incalculable.

(3) Geomorphological approaches such as the ones de-

scribed in this paper can help to improve safety factors for

dams and reservoirs at the design stage, by estimating

historic flood magnitudes to complement gauged data

series. This principle can be applicable elsewhere in the

world, and the approaches used in this area of Spain can

be used wherever evidence of past flood events is docu-

mented either in written or oral form, photographic form

or as field (geomorphological) evidence (e.g. slack-water

deposits, high water marks, etc.).

(4) However, the sources of information and dating

techniques used to extend the available data in this area

of Spain might vary in other parts of the world depending

on the available evidence and type of environment, as



there are limitations in the use of some of the techniques.

For example, trees and other organic debris associated

with high flows and flood episodes can be dated using

dendrochronology and 14C techniques, but might not be

available or suitable under conditions of regular or re-

peated slow-onset flooding. The presence or not of lichens

can be used to determine the time elapsed since the last

high-flow episode at specific locations. The existence or

not of documented or witness evidence and its quality will

determine whether this can be used to extend data series.

Furthermore, some techniques such as luminescence

dating of sediment deposits are known to work better in

semi-arid environments where the water content within

sediments is more regular for longer periods than in

highly seasonal climates such as that of the UK. This

makes luminescence techniques less suitable in temperate

environments, and indicates that a degree of good judge-

ment is required from the professional as to the selection

and adequacy of the approaches to use.

Acknowledgements

This paper would not have been possible without the

eternal support and help of my husband Yacob, to whom I

am always grateful. I also thank my PhD supervisor

Professor Phillip Ashworth, my dear friend Jose Antonio

Benavente for his help and guidance while working in

this region of Spain, and my parents for the number of

trips they have taken to the Northeast of Spain over the

years and their help in collecting field data. I am also very

grateful to Alfredo Gasion, the local photographer of Mas

de las Matas, for allowing me to use his photographs of

the 1996 flood event.

The views expressed in this paper are exclusively the

author’s views based on knowledge of the area, empirical

evidence and methods used world-wide. They do not

represent the Environment Agency’s views. The material

used in the paper includes original research by the author

constituting part of her doctoral studies.

References

Bull, W.B. (1998) Floods; Degradation and Aggradation. In

Baker, V.R., Kochel, R.C. and Patton, P.C. (eds). Flood

Geomorphology Chapter 8, pp. 157–166. New York.

Comision Tecnica de Inundaciones. (1985) Estudio de

Inundaciones Historicas: mapa de riesgos potenciales de la

Cuenca del Ebro. Vol. IV–VIII. Comision Nacional de

Proteccion Civil, Spain.

Fuller, I.C., Macklin, M.G., Passmore, D.G., Brewer, P.A.,

Lewin, J. and Wintle, A.G. (1996) Geochronologies and

Environmental Records of Quaternary Fluvial Sequences

in the Guadalope Basin, Northeast Spain, Based on

Luminescence Dating. In Branson, J., Brown, A.G. and

Gregory, K. J. (eds) Global Continental Changes: The Context of

Palaeohydrology, pp. 99–120. The Geological Society, London.

Lopez-Aviles, A. (1998) Flood regime and Quaternary

sedimentation style in the Guadalope Basin, Northeast Spain.

PhD Thesis, University of Leeds, Leeds, LS, UK.

Lopez-Aviles, A., Ashworth, P.J. and Macklin, M.G. (1998)

Floods and Quaternary Sedimentation Style in a Bedrock

Controlled Reach of the Bergantes River, Ebro Basin,

Northeast Spain. In Benito, G., Baker, V.R. and Gregory, K.J.

(eds) Palaeohydrology and Environmental Change, pp. 181–196.

Wiley and Sons, Chichester, UK.

World Commission on Dams. (2000) Dams and Development:

A New Framework for Decision Making. Earthscan, London.

World Bank. (2004) World Development Report. The World Bank,

Washington DC.



Flash flooding in Spain: geomorphological approaches supporting flood frequency analysis, and the implications for the design of structures

Documents